Dr Sikandar Shaikh, Consultant PET-CT & Radiology, Yashoda Hospitals, Hyderabad, Asst Prof Shadan Medical College Hyderabad and Adjunct Asst Prof ,Dept of Biomedical Engg. IIT Hyderabad
Artificial Intelligence (AI) is the most promising area of health innovation in medical imaging. AI may find multiple applications, from image acquisition and processing to aided reporting, follow-up planning, data storage, data mining and many others. AI is patterned after the brain’s neural networks. It uses multiple layers of non-linear processing units to ‘teach’ itself how to understand data classifying the record or making predictions.
What are the challenges of using artificial intelligence?
The principle limitation of AI is that it learns from the data. If any data is inappropriate then it has limitations.
How AI works in imaging?
The increasing amount of imaging data to be processed can influence how radiologists interpret images: from inference to merely detection and description. When too much time is taken for image analysis
- Machine learning uses methods from neural networks, statistics, operations research and physics
- A neural network is a type of machine learning that is made up of interconnected units (like neurons) that processes information by responding to external inputs, relaying information between each unit.
- Deep learning uses huge neural networks with many layers of processing units, taking advantage of advances in computing power and improved training techniques to learn complex patterns in large amounts of data.
- Cognitive computing is a subfield of AI that strives for a natural, human-like interaction with machines.
- Computer vision relies on pattern recognition and deep learning to recognise what’s in a picture or video.
- Natural language processing (NLP) is the ability of computers to analyse, understand and generate human language, including speech.
- Graphical processing units provide the heavy compute power that’s required for iterative processing. Training neural networks requires big data plus compute power.
- The Internet of Things generates massive amounts of data from connected devices, most of it unanalysed.
- Advanced algorithms are being developed and combined in new ways to analyse more data faster and at multiple levels.
- APIs, or application processing interfaces, are portable packages of code that make it possible to add AI functionality to existing products and software packages.
AI applications may enhance the reproducibility of technical protocols, improving image quality and decreasing radiation dose, decreasing MRI scanner time and optimising staffing and CT/MRI scanner utilisation, thereby reducing costs. The quicker and standardised detection of image findings has the potential to shorten reporting time and to create automated sections of reports. The reasonable doubt is that we are now facing methods that not only cover the production of medical images but also involve detection and characterisation, properly entering the diagnostic process. Indeed, this is a new challenge, but also an additional value of AI. The high number of examinations to be reported and rather focus on communication with patients and interaction with colleagues in multi-disciplinary teams.
Following are the immediate impacts of AI in Imaging
- Prioritisation of reporting
- Comparison of current and previous examinations
- Quick identification of negative studies
- Aggregation of electronic medical records
- Automatic recall and rescheduling of patients
- Immediate use of clinical decision support systems for ordering, interpreting, and defining further patient management
- Internal peer-review of reports
- Tracking of residents’ training
- Quality control of technologists’ performance and tracked communication between radiologists and technologists.
- Data mining regarding relevant issues, including radiation dose
The other possibilities with AI are
- Anticipation of the diagnosis of cancerous lesions in oncologic patients using texture analysis and other advanced approaches
- Prediction of treatment response to therapies for tumours, such as intra-arterial treatment for Hepatocellular carcinoma
- Evaluation of the biological relevance of borderline cases
- Estimation of functional parameters, such as the fractional flow reserve from CT coronary angiography using deep learning
- Detection of perfusion defects and ischemia, for example in the case of myocardial stress perfusion defects and induced ischemia
- Segmentation and shape modelling, such as brain tumour segmentation or, more generally, brain structure segmentation
- Reducing diffusion MRI data processing to a single optimised step, for example making microstructure prediction on a voxel.
Radiologists must act more as clinicians, applying their clinical knowledge in answering diagnostic questions and guiding decision-making, which represent their main tasks. They should keep their human control in the loop, considering clinical, personal and societal contexts in ways that AI alone is not able to do. Hence, an updated radiologist should be aware of the basic principles of ML/DL systems for managing this systems and can play a leading role in this oncoming change.